FedAR: Activity and Resource-Aware Federated Learning Model for\n Distributed Mobile Robots

摘要

Smartphones, autonomous vehicles, and the Internet-of-things (IoT) devices\nare considered the primary data source for a distributed network. Due to a\nrevolutionary breakthrough in internet availability and continuous improvement\nof the IoT devices capabilities, it is desirable to store data locally and\nperform computation at the edge, as opposed to share all local information with\na centralized computation agent. A recently proposed Machine Learning (ML)\nalgorithm called Federated Learning (FL) paves the path towards preserving data\nprivacy, performing distributed learning, and reducing communication overhead\nin large-scale machine learning (ML) problems. This paper proposes an FL model\nby monitoring client activities and leveraging available local computing\nresources, particularly for resource-constrained IoT devices (e.g., mobile\nrobots), to accelerate the learning process. We assign a trust score to each FL\nclient, which is updated based on the client's activities. We consider a\ndistributed mobile robot as an FL client with resource limitations either in\nmemory, bandwidth, processor, or battery life. We consider such mobile robots\nas FL clients to understand their resource-constrained behavior in a real-world\nsetting. We consider an FL client to be untrustworthy if the client infuses\nincorrect models or repeatedly gives slow responses during the FL process.\nAfter disregarding the ineffective and unreliable client, we perform local\ntraining on the selected FL clients. To further reduce the straggler issue, we\nenable an asynchronous FL mechanism by performing aggregation on the FL server\nwithout waiting for a long period to receive a particular client's response.\n